Empirical Study of Vessel Extraction Algorithms

Abstract

Medical imaging is a technique for creating an image of the human body in order to diagnose various diseases such as stenosis, aneurysm, arterial venous malformation, thrombus, plaque and internal bleeding. Blood vessel segmentation is critical in the diagnosis of a variety of diseases. Blood vessels that are segmented give much useful information about their anatomy and location. They are important in a variety of medical applications, including diagnostic, surgical therapy, and radiation treatments.  A significant amount of research has gone into vessel segmentation, and a variety of techniques has emerged as a result. In addition, there are different segmentation techniques such as active contour segmentation technique, hybrid segmentation technique, thresholding segmentation techniques, watershed segmentation techniques, edge detection segmentation technique, etc. It is also observed that magnetic resonance images of blood vessels were exposed to noise due to selection and inappropriate techniques such poor performance invisibility. In other words, there is no single approach to follow for a perfect outcome of images. There are some of the methods that use gray-level histograms, while there are others that integrate spatial image information, and this causes noisy outcomes. Therefore, we build the medical imaging vessel visualization system using MATLAB as tool. In this study, we empirically investigate the visibility performance vessel extraction algorithm. We implement following vessel extraction algorithms: active contour algorithm and edge detection algorithm. We observed that edge detection algorithm (SOBEL) is the better in term of image clarity as compared to active contour and edge detection algorithm. This project enable IS department to do more advanced level research in medical imaging.